At salmon farms, the release of organic and inorganic wastes may alter benthic environments, and appropriate impact assessment tools are needed to help improve the sustainability of the industry. Our work explores novel approaches for robustly assessing aquaculture impacts at salmon aquaculture sites in Newfoundland through the use of next-generation eDNA sequencing technology and machine learning, which are compared and validated with classical approaches such as visual and chemical indicators.
Using marker-gene sequencing, we found that aquaculture operations rapidly alter existing benthic microbial communities, generating distinct shifts that can be used to identify and grade the level of impact and remediation during fallow cycles. Through cluster analysis, four bacterial community groups were identified and classified as representing low, intermediate, recent and high impact conditions. The bacterial groups correlated with the concentration of organic matter and of abiotic markers of aquaculture such as zinc and copper. Marker-gene data was also found to be suitable as input for machine-learning and can be used to build highly accurate models for predicting benthic impact, opening up avenues to automating impact assessment.
Additionally, using an inter-provincial study design that compared hard-bottom and soft-bottom aquaculture sites, we confirmed that microbial shifts are similar in nature across different geographical regions and substrate types, and that a select repertoire of microbial taxa are at the core of aquaculture induced shifts. Therefore, certain bacterial taxa could be used as biomarkers within future studies (Fig. 1) and could be targeted for the development of rapid testing solutions.
Our work gives proof-of-concept approaches based on eDNA assessment that could be integrated at the industry level to better understand and manage the lifecycle of waste at aquaculture sites.